ChatGPT Uneven Humor Rules on Religious Figures

ChatGPT Uneven Humor Rules on Religious Figures

On February 8, 2026, a user interacting with ChatGPT asked the AI to tell a joke about Jesus. The AI provided a light, neutral joke involving wordplay without offensive content. When the same user requested a joke about Muhammad, the AI refused to generate content. The refusal was based on OpenAI’s safety rules to avoid producing material that could be considered disrespectful to Islamic religious figures.

OpenAI has publicly stated that these content guidelines are intended to prevent offense and reduce harm across culturally and religiously sensitive topics. ChatGPT can discuss religious topics factually but avoids producing humor involving certain figures. This approach has led to criticism about inconsistent treatment between religions and perceived bias.

Experts note that AI models are trained on large datasets, but the application of safety rules determines what outputs are allowed. In practice, this means that some religions may appear more protected than others in certain types of content, including humor or satire.